Modelling of daily PM₁₀ concetrations using Markov chain model at selected area in peninsular Malaysia / Norsalwani Mohamad

The purposes of this study are to investigate the occurrences of particulate matter with aerodynamic diameter less than or equal to 10 μm (PM10) concentration and develop the PM10 concentration index (PCI). There has been very little work in predicting the sequence of PM10 concentration. This study considers the use of the Markov chain model as it has advantages due to the dependency of the previous events and being highly suitable for the pattern of observations. Twelve years of daily PM10 concentration data (2002-2013) at three monitoring stations in Peninsular Malaysia were used in this study. The assumption of the Markov chain model was met when the data used possessed the Markov chain properties as successive events that depended on each other. The results showed that the higher order was more appropriate for the monitoring stations with the threshold value less than 100 μgm-3 for both decision criteria (AIC and BIC). Based on the optimum order, the occurrence of polluted (or non-polluted) days was found to be depended on the condition of two or three days before the observed day where the prediction of PM10 events can be made based on the two or three days before the observed day. However, up to four orders were suggested as the use of the higher order was less practical due to the increasing number of parameters and difficulties to estimate the parameters. Thus, it can be concluded that at least a three-day event before the PM10 concentration event is needed to minimise the effect and better precautions can be taken. The proposed PCI describes the severity of exposure impact proneness (EIP) due to the pollutant persistency level in environment. The results show that Shah Alam recorded the highest value of API (PM10) in “Critical EIP” condition followed by Pasir Gudang and Kuala Terengganu.